Studies Towards the Total Synthesis of (+)- Retronecine ...

of 214/214
Studies Towards the Total Synthesis of (+)- Retronecine and Anthracimycin Joshua C. Smith Doctor of Philosophy University of York Chemistry September 2016
  • date post

    11-Jan-2022
  • Category

    Documents

  • view

    1
  • download

    0

Embed Size (px)

Transcript of Studies Towards the Total Synthesis of (+)- Retronecine ...

Retronecine and Anthracimycin
Joshua C. Smith
Doctor of Philosophy
University of York
2
Abstract
This thesis consists of two separate projects. The first project involved an attempted
total synthesis of (+)-retronecine A, a poisonous pyrrolizidine alkaloid via a novel
asymmetric α-lithiation-substitution reaction of N-thiopivaloyl azetidine B, mediated by
a chiral diamine.
A total synthesis was not possible and our attention turned towards elucidating the
mechanism of enantioinduction. Of particular note was how the use of carbon dioxide
and methyl chloroformate as electrophiles gave products with the opposite
configuration under otherwise identical reaction conditions (using (–)-sparteine C).
Chapter two consists of work undertaken towards a total synthesis of anthracimycin D,
a potent marine antibiotic. The development of robust methodology was explored for
the early stages of the synthesis. In addition, a synthesis of the model decalinone E, was
developed, via a Sakurai-aldol reaction, followed by ring closing metathesis.
3
Acknowledgements..........................................................................................................9
Author's Declaration.....................................................................................................10
Chapter One: Studies Towards the Total Synthesis of (+)-Retronecine via an
Azetidine Lithiation-Trapping Reaction.....................................................................11
1.2 α-Lithiation-Substitution of Nitrogen Containing Heterocycles...................12
1.2.1 α-Lithiation-Substitution of 5-Membered Nitrogen Containing
Heterocycles............................................................................................14
Heterocycles………………………………………………………….21
Heterocycles…………………………………………………………....23
Heterocycles…………………………………………………………..27
1.3 Project Outline...............................................................................................35
1.5 Asymmetric α-Lithiation Substitution Reactions of N-Thiopivaloyl Azetidine
and Pyrrolidine....................................................................................................57
4
N-Thiopivaloyl Azetidine and Pyrrolidine..............................................59
Chapter Two: Studies Towards the Total Synthesis of Anthracimycin………...…71
2.1 Introduction to Antibiotic Resistance……………………………………..71
2.2 Introduction to Anthracimycin And Chlorotonil…………………………...72
2.3 Project Outline...............................................................................................79
2.4.2 Oxidation of Ketone 139................................................................86
2.5 Investigation of Cuprate Additions................................................................91
2.6 Synthesis of the Decalin Framework of Anthracimycin.............................110
2.6.1 TiCl4 Mediated Addition of Allyltrimethylsilane to Cyclohexenone..
………………………………………………………………...………110
2.6.2 Aldol and Ring Closing Metathesis Route to Decalins………....120
2.6.3 Proof of Relative Stereochemistry of Aldol Products 188 and 189...
…………………………………………………………………...……131
Chapter Three: Experimental Procedures................................................................135
5
List of Abbreviations...................................................................................................204
Table 1.2: Results Demonstrating Thermodynamic and Kinetic Resolutions of 8.........19
Table 1.3: Yields Of 1,2-Migration Products on a Variety of N-Boc Aziridines………23
Table 1.4: Lithiation-Boronate Rearrangements of Isopropyl N-Boc Aziridine.............24
Table 1.5: (–)-1 Mediated Lithiation-Trapping of Borane-Aziridine Complexes...........27
Table 1:6 Optimisation of the Asymmetric Lithiation-Trapping of 35………………...30
Table 1.7: Results of Self-Addition Reactions of 2-Aryl N-Boc Azetidines…………...34
Table 1.8: Role of Chiral Diamine in the Lithiation-Trapping of 35 with Benzaldehyde..
………………………………………………………………………....................…….57
Table 2.2: IBS Mediated Oxidation of 4-Substituted Cyclohexanols.............................89
Table 2.3: Failed 1,4-Additions of Allyl Cuprate to 117...............................................100
Table 2.4: Optimisation of Conditions for 1,4-Addition of Allyl Cuprate to 117…….103
Table 2.5: Optimisation of Conditions for the Iodination of 117…………………….107
Table 2.6: Alkylation of Titanium Enolates with Ethylene Oxide................................113
Table 2.7: Reactions of 180 with Isobutylene Oxide....................................................120
7
9
Acknowledgements
As a joint student of two supervisors, I feel fortunate to have been afforded twice the
input, ideas and expertise as I would have received were I only a student of one. It has
been a privilege to have known and worked with Dr Clarke and Professor O'Brien and
despite my occasional brashness, I have the utmost respect for them both.
I would like to thank all members of the PAC and POB groups for their input and
suggestions over the years, in particular to Kristaps Ermanis for showing me the ropes
in my early days of my PhD, as well as to Peter Rayner for introducing me to the
practicalities of organolithium chemistry and his shared experiences of banging one's
head against a wall. I would also like to thank the PAC group for the daily banter which,
sometimes savage, often silly but always funny, has made day to day life in the lab that
bit brighter. Long may the wall of Sam live on.
As is common to all PhD students, this work would not have been possible without a
great deal of support from technical staff in the department. I would therefore like to
extend my thanks towards Heather Fish for the NMR service, as well as her own unique
brand of humour, to Karl Heaton for mass spectrometry, Adrian Whitwood for X-ray
services and Steve and Mike from stores and the leeway they afforded myself and the
PAC group.
I would like to thank my entire family, in particular my mum and her husband John,
without whose love, support and patience this PhD may well never have reached
completion. I would also like to make mention of the people I have met at York
Vineyard, their friendship and hospitality has been greatly appreciated and I consider all
of them to be my extended family.
Lastly but by no means least, I would like to give thanks to our Lord Jesus Christ for the
grace and mercy He has shown to me each day, who has brought me through some
difficult circumstances and blessed me with opportunities, friendships and success that I
can scarcely believe nor deserve. May what little contained herein be for His glory, who
is above all others and blessed forever, Amen.
10
Author's Declaration
The research and results presented herein are, to the best of my knowledge, original
except for when due reference has been made to other authors and/or co-workers.
Some of the work in this thesis has been published in the following paper:
“Mechanistic interrogation of the asymmetric lithiation-trapping of N-thiopivaloyl
azetidine and pyrrolidine.”
Chem. Commun., 2016, 52, 1354-1357
This thesis has not been submitted for any other award at this or any other institution.
11
via an Azetidine Lithiation-Trapping Reaction
1.1 Introduction to Nitrogen Containing Natural Products
Natural products, assuming that they are chiral, are usually produced as a single
enantiomer, given that they are typically made via enzymatic catalysis. Despite major
advances, efficient methods for asymmetric synthesis remains a challenge in natural
product total synthesis, in terms of matching the enantiomeric excesses achieved by the
organism that produces it, stepwise efficiency and substrate specificity.
Nitrogen heterocycles are biologically privileged motifs. Many natural products contain
at least one substituted nitrogen heterocycle and some selected examples are shown in
figure 1.1, including Miraziridine A, a protease inhibitor, 1 Desoxyprosopine, an alkaloid
isolated from Prosopsis africana, 2
Penazetidine A, a protein kinase C inhibitor, 3 and
(–)-Kainic acid, a mammalian neuroexcitory amino acid. 4
Figure 1.1
Methods for the asymmetric synthesis of α-substituted nitrogen heterocycles are thus
important for the synthetic chemist. To date, the asymmetric synthesis of functionalised
organolithium reagents remains one of the best, most conceptually simple and fastest
ways of accessing enantioenriched α-substituted nitrogen heterocycles.
12
Typically, organolithium reagents are formed via three methods: reductive lithiation,
transmetallation and deprotonation, summarised in Scheme 1.1.
Scheme 1.1
Ultimately, reductive lithiation remains the only method of generating organolithium
reagents ab initio but, due to the single electron transfer mechanism of their formation 5
which proceeds via an sp 2 planar radical intermediate, any stereogenic centres
containing a C–X bond are racemised. Transmetallation also generally suffers from the
same problem, although tin-lithium exchange provides a method for retaining the
enantioenrichment of the starting stannane. 6 Deprotonation, therefore, remains the most
general and most effective method for the formation of enantioenriched α-functionalised
organolithium reagents.
The most well utilised method of carrying out asymmetric deprotonations of a substrate
is to use a strong base (sBuLi or nBuLi typically) in the presence of a chiral diamine,
with (–)-sparteine (–)-1, an alkaloid extracted from Cytisus scoparius, 7 being the prime
example. Although (–)-sparteine (–)-1 was used in organometallic chemistry as early as
1968, 8 its potential for asymmetric deprotonations was not realised until 1990, when
Hoppe 9 treated cyclic carbamate 2 with sBuLi in the presence of (–)-sparteine to form
enantioenriched carbamate 3, which could then be trapped with a variety of
electrophiles in high yields and enantioselectivities. The synthesis of α-functionalised
carbamates (R)-4 and (S)-5 are shown in Scheme 1.2.
13
Scheme 1.2
Importantly, the carbamate functional group acts as a directing group and coordinates to
the sBuLi, forming a pre-lithiation complex 6 consisting of diamine, sBuLi and the
starting carbamate 2. Complex 6 can then undergo deprotonation to give the
functionalised organolithium. Known as the "complex induced proximity effect",
evidence for the pre-lithiation complex was first put forwards by Beak 10
in 1995 and has
subsequently been directly observed with some N-Boc substrates by in situ IR
spectroscopy. 11, 12
It is also worth noting, that chiral base mixtures consisting of an
alkyllithium and (–)-1 require a non-coordinating solvent such as Et2O, toluene or
alkanes in order for effective enantioinduction to occur. If THF is used, the product of
lithiation-trapping is usually racemic 13-20
which is believed to be due to the solvent out-
competing (–)-1 for ligation to the lithium. 13, 21
14
Perhaps the most well utilised substrate for α-lithiation-trapping chemistry is the cyclic
N-Boc pyrrolidine 7. First reported in 1991, 22
7 was treated with sBuLi in the presence
of (–)-1 to form enantioenriched organolithium (S)-8 and was then trapped with a
variety of electrophiles. Some selected examples are shown in Scheme 1.3.
Scheme 1.3
Crucially, Beak also demonstrated that the mechanism of enantioinduction is an
enantioselective deprotonation of 7 by the sBuLi/(–)-1 complex, rather than a resolution
of racemic organolithium 8 and (–)-1. The key experiments are shown in Table 1.1.
Enantioenriched stannane (S)-9 (97:3 er) was transmetallated to the corresponding
organolithium (S)-8 in the presence of TMEDA and then trapped with Me3SiCl. The
product silane had an enantiomeric ratio of 83:17 (Table 1.1, entry 1). In contrast, when
racemic 9 was transmetallated to (S)-8 in the presence of (–)-1, then trapped with
Me3SiCl, the corresponding silane was effectively racemic (Table 1.1, entry 2).
15
Table 1.1: Demonstration of Configurational Stability of 8
The results indicate that organolithium 8 is configurationally stable under the reaction
conditions ie., it is not able to interconvert to its enantiomer. This means that both the
sense of induction and the enantioenrichment of the product must be determined during
the deprotonation step (Scheme 1.4). This also explains why the sense of induction does
not change and why the overall enantioselectivities do not change significantly with the
electrophile that is used.
opposite enantiomeric series to (S)-proline, (+)-sparteine (+)-1 has been less available
commercially. To date, the only enantioselective synthesis of (+)-1 was reported by
Aubé 23
using norbornadione (S,S)-11 as starting material. This route gave (+)-sparteine
(+)-1 in 15 steps and 15.7% overall yield (Scheme 1.5).
Scheme 1.5
For this reason, chiral diamines that give the opposite sense of induction to (–)-1 have
been investigated. Originally developed as ligands to facilitate the asymmetric addition
of organolithium reagents to aromatic imines, 24, 25
diamines (S,S)-12 and (R,R)-12 have
been shown to be effective ligands in the asymmetric α-lithiation-substitution of 7. 26
For
example, N-Boc pyrrolidine 7 was treated with sBuLi in the presence of (R,R)-12 and
trapped with Me3SiCl to give silane (S)-10 in 72% and 95:5 er (Scheme 1.6), comparing
favourably to the enantioselectivities obtained with (–)-1.
Scheme 1.6
Crucially, racemic trans-1,2-diaminocyclohexane is commercially available. Thus,
either enantiomer of the desired chiral diamine can be synthesised following resolution
with (D)-tartaric acid. (R,R)-12 was synthesised via two successive alkylations in 60%
overall yield. 27, 28
(–)-1 derived from (–)-cytisine (–)-14, developed within our group, 29
has shown the
greatest promise in terms of matching the yields and enantioselectivities obtained with
(–)-1. The synthesis of (+)-13 and a comparison of its use in α-lithiation trapping are
shown in Scheme 1.7.
Scheme 1.7
The sBuLi/(+)-13 complex has been demonstrated, via a competition experiment, to be
significantly more reactive than sBuLi/(–)-1 (Scheme 1.8). 30
Given that the sBuLi/(+)-
13 complex and the sBuLi/(–)-1 complex give products with the opposite sense of
induction but at similar levels of enantioenrichment, the rate at which each complex
lithiates N-Boc pyrrolidine 7 will be proportional to the enantioenrichment of the final
product. In this case, Me3SiCl as electrophile gave (R)-10 in 62% yield and an
enantiomeric ratio of 90:10, indicating that the sBuLi/(+)-13 complex lithiates N-Boc
pyrrolidine 7 approximately twenty times faster than does the sBuLi/(–)-1 complex.
18
13 C NMR spectroscopy carried out within our
group found that the (+)-sparteine surrogate (+)-13 forms a monomeric iPrLi/diamine
species 15 (Figure 1.2) in THF. 31
To the best of our knowledge, no other monomeric
organolithium species has been directly observed.
Figure 1.2
Naturally, being able to predictably access both enantiomers of α-substituted
pyrrolidines is synthetically useful and both (–)-sparteine (–)-1 and the (+)-sparteine
surrogate (+)-13 have been used in this manner in several total syntheses. 32-34
Lithiated N-Boc pyrrolidine 8 is known to become configurationally unstable at
temperatures above –40 °C. 35
However, provided that the temperature of the reaction
remains below this, the mechanism of enantioinduction must remain an enantioselective
deprotonation as shown in Scheme 1.4. However, if the temperature is allowed to
increase beyond –40 °C, the ratio between the enantiomers of the lithiated intermediate
will be dependent on the equilibrium constant, Keq. In 2006, Coldham 35
formed the
racemic organolithium 8 at –78 °C and allowed a resolution by warming the reaction
to –20 °C in the presence of (S,S)-16, then re-cooling to –78 °C and trapping with
Me3SiCl (Table 1.2).
Table 1.2: Results Demonstrating Thermodynamic and Kinetic Resolutions of 8
The results indicate that chiral diamine (S,S)-16 favours a 58:42 thermodynamic ratio of
enantiomers, with the (R)-lithio-pyrrolidine 8 being the major diastereoisomer. As
lithiated pyrrolidine 8 is configurationally stable at –78 °C, the addition of excess
Me3SiCl preserves the thermodynamic ratio of (R)-8 and (S)-8 in the enantiomeric ratio
of the product (Scheme 1.9).
Scheme 1.9
20
Conversely, the addition of substoichometric amounts of Me3SiCl gives the silane (S)-
10 as the major product, with higher enantioselectivities but lower yields. This implies
that the (S)-lithiopyrrolidine (S)-8/(S,S)-16 complex reacts faster with Me3SiCl than its
diastereoisomer. Coldham was able to optimise the reaction conditions to give silane
(S)-10 in 57% yield and 95:5 er by the slow addition of excess Me3SiCl at –20 °C
(Scheme 1.10). The presence of 10 equivalents of nBuLi was required to maximise ers,
possibly to increase the rate of interconversion of (S)-8 and (R)-8.
Scheme 1.10
Nonetheless, resolutions of lithiated N-Boc pyrrolidine 8 often lack generality due to the
poor selectivities offered by thermodynamic resolutions, whilst kinetic resolutions are
highly electrophile dependent, both for enantioselectivities and for the overall sense of
induction. In contrast, enantioselective deprotonations always proceed with the same
sense of induction and enantioselectivities, provided that trapping occurs below the
temperature at which the organolithium becomes configurationally unstable.
21
Methodology developed for the asymmetric α-lithiation substitution of N-Boc
pyrrolidine 7 does not produce equivalent results when applied to the 6-membered ring
equivalent N-Boc piperidine 17. Although N-Boc pyrrolidine 7 fully lithiates at –30 °C
in THF in 5 min, even in the absence of a diamine, N-Boc piperidine 17 does not lithiate
at all under the same reaction conditions. 36
Even in the presence of (–)-sparteine (–)-1,
N-Boc piperidine 17 only undergoes 10% deprotonation at –78 °C after 6 h using
sBuLi. 12
It is known that in the case of N-Boc pyrrolidine 7 and N-Boc piperidine 17,
the equatorial proton is preferentially removed, irrespective of whether a chiral ligand is
present. 37, 38
When (–)-1 is used the chiral diamine, the kinetic energy barrier to
deprotonation of the pro-(S) proton from 17 is lower than for the pro-(R) proton. 39-41
In
contrast to (–)-1, the use of the (+)-sparteine surrogate (+)-13 leads to complete
deprotonation over 2 h at –78 °C, as confirmed by in situ IR spectroscopy. 12
Enantioselectivities are similar to those obtained with (–)-1 but with the expected
opposite sense of induction (Scheme 1.11). 12, 42
As stated previously, (+)-13 has been
shown to form a monomeric iPrLi/diamine complex in THF. As lower order
alkyllithium species are typically more reactive than their higher order counterparts, 43
it
may simply be that due to the increased binding affinity of (+)-13 relative to (–)-1, it is
better able to deaggregate the sBuLi to a more reactive species that can more readily
overcome the kinetic energy barrier of deprotonation.
Scheme 1.11
Despite the difference in relative kinetic barriers, lithiated N-Boc piperidine 18 has been
demonstrated to be configurationally stable at –78 °C. 12
The mechanism of
enantioinduction, provided the temperature is kept at –78 °C, must therefore be an
enantioselective deprotonation as is the case with N-Boc pyrrolidine 7. This was
formally proven by tin-lithium exchange at –78 °C on racemic stannane 19. After tin-
lithium exchange, (+)-1 was added and the reaction was incubated at –78 °C for 2 h.
22
The lithiated intermediate was then trapped with methyl chloroformate to give rac-20,
indicating that enantioinduction does not occur after the deprotonation event (Scheme
1.12).
Scheme 1.12
In contrast to N-Boc piperidine 17, N-Boc piperazine 21 lithiates rapidly, even in the
absence of a diamine (Scheme 1.13). 36
Although the exact reason for this effect is
unknown, two possible explanations for this effect can be suggested. It could be that N-
Boc piperazine 21 adopts a different conformation to N-Boc piperidine 17 and thus
allows more facile deprotonation. Alternatively, the electron withdrawing β-nitrogen
could lower the pKa of the α-protons in N-Boc piperazine 21.
Scheme 1.13
Although α-lithiation-substitution chemistry is well established for N-Boc pyrrolidine 7
and N-Boc piperidine 17, N-Boc aziridines behave differently under similar conditions.
In 1994, Beak carried out the first α-lithiation-substitution reaction of N-Boc aziridine
22 and the 2-methyl 23 and 2-trimethylsilyl derivatives 24 (Scheme 1.14). 44
Scheme 1.14
Beak did not comment on the diastereoselectivity of the reaction, although it is likely
that the two ring substituents are trans to one another, as has been demonstrated on
similar substrates. 45-48
It is also worth noting that the substituted aziridine could only be
accessed with an in situ Me3SiCl trap. In the original reporrt, Beak did not speculate on
why this was the case, although it is now known that 2-lithio-N-Boc aziridines rapidly
undergo a 1,2-migration to give the corresponding 2-substituted free amine, as
demonstrated by Hodgson (Table 1.3). 47
Table 1.3: Yields of 1,2-Migration Products on a Variety of N-Boc Aziridines
24
It is unclear why this migration occurs with N-Boc aziridines but not with larger ring
sizes. A possible reason may be that the nitrogen lone pair does not delocalise into the
π* C=O orbital as efficiently in three-membered ring (Figure 1.3).
Figure 1.3
The extra ring strain of N-Boc aziridine 22 has been exploited by Aggarwal 46
to
The reaction proceeds diastereoselectively and enantioenriched starting materials retain
their original enantioenrichment in the product (Sheme 1.15).
Scheme 1.15
The tendency of the Boc group to migrate largely restricts the applicability of more
traditional α-lithiation-substitution chemistry to unreactive electrophiles that can be
used for in situ trapping ie. Me3SiCl and boronates.
25
reported the first successful directed α-lithiation-substitution
reaction on racemic aziridine 25 without using utilising an in situ electrophilic trap. The
Boc group was substituted for a t-butyl sulfone directing group, thus avoiding the
previously unwanted side reaction of 1,2-migration (Scheme 1.16).
Scheme 1.16
Lithiation is both regioselective and diastereoselective, with the proton trans to the alkyl
chain being removed and is compatible with a wide range of electrophiles. However, to
the best of our knowledge, no asymmetric variant of this reaction has been reported.
In 1997, Vedejs 45
reported the first example of an α-lithiation-substitution reaction on
an aziridine without an internal directing group. TBS protected aziridine 26 was treated
with borane to give the stable aziridine-borane complex 27 which could then be lithiated
with sBuLi and trapped with a variety of electrophiles (Scheme 1.17).
26
Scheme 1.17
Lithiation occurs syn to the boron, although it is likely that this is purely down to sterics,
rather than any directing effect from the boron. Crucially, Vedejs demonstrated that the
lithiated intermediate is configurationally stable at –78 °C. Tin-lithium exchange on a
77:23 diastereomeric mixture of stannanes rac-28 and rac-29, followed by trapping
with Bu3SnCl gave re-formed stannanes rac-28 and rac-29 in a largely unchanged
82:18 diastereomeric ratio (Scheme 1.18). When aziridine 30 was lithiated and trapped
with Bu3SnCl, stannanes rac-28 and rac-29 were formed in a 98:2 diasteremeric ratio
(Scheme 1.18). The results imply that diastereoselectivity is not due to a resolution of
the 2-lithio borane-aziridine, otherwise both experiments would have given an identical
diastereomeric ratio.
Scheme 1.18
reported the first asymmetric α-lithiation substitution reaction on a
veriety of aziridine-boranes. The use of (–)-sparteine (–)-1 as the chiral diamine gave
access to 2-substituted aziridines in a consistent er of approximately 85:15, although the
absolute configuration of the products was not assigned (Table 1.5).
Table 1.5: (–)-1 Mediated Lithiation-Trapping of Borane-Aziridine Complexes
1.2.4 α-Lithiation-Substitution of 4-Membered Nitrogen Containing Heterocycles
In contrast with 3-,5- and 6-membered rings, lithiation-trapping of the corresponding 4-
membered rings, azetidines, has not been as widely explored. Prior to 2010, the only
known examples of successful lithiation on an azetidine were reported by Seebach 51, 52
in 1977 and 1981. Azetidines 31 and 32 were lithiated with tBuLi and LDA,
respectively, and then trapped with benzaldehyde to give product alcohols rac-33 and
rac-34 in 62% and 65% yield (Scheme 1.19). The relative stereochemistry of rac-33
and rac-34 was not proven and the diastereoselectivity was not mentioned in either
instance. The harsh conditions required to lithiate triphenyl acyl azetidine 31 also mean
it is unclear as to whether the carbonyl group acts as a true directing group in the
deprotonation step, or whether it is simply due to the electron withdrawing effect of the
nitrogen atom.
substitution reaction on azetidine 35 (Scheme 1.20).
Scheme 1.20
Yields were good with all the electrophiles tested. Interestingly, using benzaldehyde as
the electrophile, Hodgson reported that rac-38 was formed as a single diastereoisomer.
Of greater note is that the use of Boc, SOtBu, SO2tBu and PO(OEt)2 as directing groups
either led to decomposition, only partial lithiation or no observable lithiation. Of the
protecting groups tested, only the thiopivalamide, first reported by Seebach 53
in 1976
(Scheme 1.21), allowed complete lithiation to occur at the α position.
29
Scheme 1.21
Use of the thioamide directing group may allow facile deprotonation due to both the
longer C=S bond length relative to the C=O bond length and the larger, more diffuse sp 2
p-orbitals containing the lone pairs on the sulfur atom (Figure 1.4). This may allow the
reactive carbon centre of sBuLi to come into closer proximity to the α-protons and
permit efficient deprotonation. 54, 55
Figure 1.4
The thioamide group is less synthetically useful than the Boc group. However, some
useful transformations are possible (Scheme 1.22). For example, the thioamide can be
removed with MeLi or refluxing ethylenediamine, oxidised to the corresponding amide
using hydrogen peroxide or desulfurised by treatment with LiAlH4.
30
Hodgson also reported the first example of an asymmetric α-lithiation substitution of N-
thiopivaloyl azetidine 35 (Table 1.6). Of the four ligands tested, (R,R)-12 gave the
highest enantioselectivity, forming methylated (R)-40 in an 80:20 enantiomeric ratio
(Table 1.6, entry 4). The results were intriguing. For example, (–)-1 which works well
with N-Boc heterocycles gave low enantioselectivities and no explanation was offered
on the mechanism of enantioinduction.
Table 1.6: Optimisation of the Asymmetric Lithiation-Trapping of 35
31
Hodgson 56
has also utilised the thiopivaloyl moiety to lithiate azetidinol 42, under
similar conditions used to lithiate 35 (scheme 1.23). For all except deuterium, the
electrophile takes the trans-position to the hydroxyl group.
Scheme 1.23
has also reported the use of the t-butoxythiocarbonyl group
(Botc) to direct lithiation to the α-position of azetidine (Scheme 1.24).
Scheme 1.24
observed with methylated azetidines 40 and 44 (Scheme 1.25). 58
Lithiation-trapping of
azetidine 40 gives rise to gem-dimethyl compound 45, whereas the analogous reaction
on azetidine 44 gives rise to the 2,4-dimethyl compounds rac-46 and rac-47.
32
Scheme 1.25
For 40, the disfavourable steric interaction between the t-butyl and methyl groups would
be an obvious explanation as to why the trans rotamer is favoured (Scheme 1.26). For
44, the additional C-O bond increases the distance between the t-butyl and methyl
groups and would thus decrease the disfavourable interaction, although why the cis
isomer itself is favoured is unclear.
Scheme 1.26
As with azetidine 35, 43 proved amenable to asymmetric lithiation-trapping with
diamine (S,S,S,S)-48, another diamine initially developed by Alexakis 59
in 2012,
proving to be the best ligand, with a maximum er of 92:8 obtained with acetone as the
electrophile (Scheme 1.27).
Although Hodgson has reported the first asymmetric α-lithiation-substitution reactions
on 35 and 43, no explanation was offered as to how enantioinduction occurs.
Recently, Luisi 60
investigated the behaviour of various lithiated 2-aryl-N-Boc azetidines
(Table 1.7). Given that N-Boc azetidine was previously proven to be resistant to
lithiation by Hodgson, the aromatic ring must therefore significantly acidify the α-
hydrogens, most likely via delocalisation of the negative charge into the aromatic ring.
Treatment of a variety of 4-substituted aromatic N-Boc azetidines with sBuLi and
trapping with acetone gave the self-addition products shown below in Table 1.7.
Interestingly, when ortho-tolyl-N-Boc azetidine was lithiated, the corresponding self
addition products were not observed (Scheme 1.28).
34
Table 1.7: Results of Self-Addition Reactions of 2-Aryl N-Boc Azetidines
Scheme 1.28
1.3 Project Outline
The starting point in this project was the development of an effective lithiation-trapping
based synthetic route for the total synthesis of (+)-retronecine (+)-49 (Figure 1.5), a
hepatotoxic pyrrolizidine alkaloid found in Jacobaea vulgaris. 61
Figure 1.5
The earliest total synthesis of racemic retronecine was reported in 1962 by Geissman. 62
Starting from carbamate-ester 50, racemic retronecine was synthesised 0.91% yield and
12 steps (Scheme 1.29).
syntheses, 63-68
our proposed synthesis would be only eight steps, comparing favourably
to the most recent synthesis by Roche who synthesised (+)-retronecine in 13 steps. Our
proposed synthesis would synthesise azetidine 35 by a two step acylation-thiolation
reported by Hodgson. 20
This would be followed by an asymmetric lithiation-trapping
using an aldehyde such as 51 as the electrophile. If azetidine 35 behaves in a similar
manner to N-Boc pyrrolidine 7 or N-Boc piperidine 17, then the (+)-sparteine surrogate
(+)–13 should give the product with the correct configuration. Hodgson also reported
that when benzaldehyde was used as electrophile, the reaction was completely
diastereoselective in favour of the syn diastereoisomer 38. Our proposed synthesis
would make use of this diastereoselectivity to control two stereocentres in one step to
form (S,S)-52 (Scheme 1.30).
From then, methyllithium-mediated deprotection would be followed by alkylation with
allyl bromide to give azetidine (S,S)-53, then mesylation and ring expansion would be
carried out to give pyrrolidine (R,R)-54. This reaction proceeds by initial mesylation of
the alcohol, followed by an intramolecular nucleophilic substitution of the mesylate by
the nitrogen lone pair, generating a transient aziridinium species. 69
This would then be
ring opened in another nucleophilic substitution reaction with a hydroxide anion to
generate the product pyrrolidine (Scheme 1.31). Ring closing metathesis and fluoride
deprotection should finally afford (+)-retronecine (+)-49 in eight linear steps.
37
Scheme 1.31
In order to carry out an effective synthesis of (+)-retronecine, several mechanistic
aspects of the asymmetric α-lithiation-trapping of N-thiopivaloyl azetidine 35 needed to
be explored. Chapter 1.4 will explore racemic lithiation-trapping of N-thiopivaloyl
azetidine and the unexpected reactivity of the lithiated intermediate. Chapter 1.5 will
then explore the mechanism of asymmetric induction to determine whether
enantioinduction occurs via an enantioselective deprotonation or a dynamic resolution.
38
To begin with, azetidine thiopivalamide 35 was prepared from azetidine hydrochloride
according to the procedure reported by Hodgson. 20
The two-step acylation-thiolation
reaction sequence afforded N-thiopivaloyl azetidine 35 in an overall yield of 77%
(Scheme 1.32). Both the intermediate amide 56 and the thioamide 35 could be purified
by vacuum distillation and the reaction sequence worked reliably on a multi-gram scale.
Scheme 1.32
With our ultimate goal being a total synthesis of (+)-retronecine (+)-49, understanding
the behaviour of the azetidine lithiation-trapping reaction was important, especially for
acrolein-type electrophiles which had not been explored previously. Therefore, our first
α-lithiation substitution reactions on thioamide 35 were carried out under racemic
conditions using sBuLi and TMEDA as diamine and benzaldehyde as the electrophilic
trapping partner. This was done to explore comparability with Hodgson's results and to
gain familiarity with this chemistry. Thus lithiation-substitution with benzaldehyde as
electrophile gave a 65:35 mixture of diastereomeric alcohols syn-38 and anti-57 by ( 1 H
NMR spectroscopy of the crude reaction mixture). They were readily separated by flash
column chromatography to give syn-38 in 37% yield and anti-57 in 19% yield (Scheme
1.33).
Scheme 1.33
The identity and relative stereochemistry of syn-38 and anti-57 were confirmed by X-
ray diffraction on a suitable single crystal of each diastereoisomer (Figure 1.6).
39
the alcohol syn-38 (Scheme 1.34).
Scheme 1.34
However, in our hands, anti-58 was always observed, even after multiple repeats of the
experiment. Upon closer inspection of the supporting information of Hodgson's paper, it
is clear that anti-58 was present in the 1 H NMR spectrum of syn-38. Yields of the
reaction in our hands were also noticeably lower than that reported by Hodgson.
Although the major diastereoisomer (syn-38) was the same, the presence of the minor
diastereoisomer (anti-58), coupled with the lower than anticipated yields led to a need
to develop more robust synthetic methodology before attempting reactions with
40
protected aldehyde 51 as needed for the total synthesis of (+)-retronecine (+)-49.
Acrolein possesses both aldehyde and alkene functionality, but is commercially
available and would serve as a reasonable model substrate in lieu of aldehyde 51. Thus
azetidine thiopivalamide 35 was subjected to α-lithiation-substitution using acrolein as
the electrophile (Scheme 1.35).
Scheme 1.35
Diastereomeric alcohols syn-58 and anti-59 were formed in a 60:40 diastereomeric ratio
(by 1 H NMR of the crude reaction mixture) in a disappointing 38% combined yield. The
relative stereochemistry of each alcohol was determined by separately acylating each
alcohol with para-nitrobenzoyl chloride to form the solid para-nitrobenzoate esters,
then crystallising and subjecting each to X-ray diffraction (Scheme 1.36). The X-ray
crystal structures are shown below in Figure 1.7.
Scheme 1.36
Figure 1.7
Given that neither benzaldehyde nor acrolein possess enolisable protons, it was
envisaged that the cause of the low yields would not be due to the trapping step, but
rather due to instability of the lithiated intermediate. Hence, we focused our attention on
the lithiation step of the reaction and investigated whether or not lithiated azetidine 35
was stable under the reaction conditions.
Thus, azetidine 35 was lithiated at –78 °C in THF, allowed to stir for 1 h and then
quenched with methanol (Scheme 1.37). Ideally, 35 should lithiate cleanly to give the
lithiated azetidine. If the organolithium is stable, then starting material 35 should be
recovered in quantitative yield upon addition of MeOH.
42
Scheme 1.37
From this experiment, thioamide 35 was recovered in only 46% yield, along with a
significant quantity of an unknown product 62. Based on a HRMS m/z value of
258.1345 for the (M+H) + peak for the unknown product, we calculated a 22% yield. It is
also worth noting, that there is still 32% of material that was unaccounted for during
this reaction, implying that there is an unknown deleterious reaction pathway.
Unfortunately, we have not been able to definitively confirm the structure of the
unknown product 62. It is useful however, to present the analytical data here and our
speculations. The 1 H NMR,
13 C NMR and DEPT spectra and accompanying data are
shown in Figures 1.8, 1.9 and 1.10.
Figure 1.8
Data for unknown product 62:
1 H NMR (400 MHz, CDCl3) δ 5.01 (dd, J = 7.0, 3.5 Hz, 1H), 2.97 (ddd, J = 11.0, 10.0,
5.0 Hz, 1H), 2.87–2.81 (m, 1H), 2.48–2.31 (m, 2H), 1.22 (s, 9H), 1.12 (s, 9H)
13 C NMR (100.6 MHz, CDCl3) δ 179.2, 93.9, 86.0, 40.3, 39.8, 38.2, 31.9, 29.4, 27.5
From the data, the presence of two t-butyl groups was clearly seen, each signal in the 1 H
NMR spectrum integrating for 9H. The presence of the azetidine ring hydrogens,
integrating to a total of 5H, is also clearly observed. Initially, we believed this unknown
product 62 to be self-addition product 63 (Figure 1.11), akin to what has also been
independently observed by Luisi. 60
Figure 1.11
We believed that this self-addition product would be formed by attack of lithiated
azetidine 64 on unlithiated starting material, followed by collapse of the tetrahedral
intermediate upon quenching, expelling a molecule of azetidine and forming self-
addition product 63 (Scheme 1.38).
Scheme 1.38
As previously stated, HRMS gave an m/z value of 258.1354, as expected for self-
addition product 63 and therefore appeared to confirm our initial assignment. However,
the 13
C NMR spectrum showed the absence of both C=S groups. Thioketones are
known to have higher 13
C chemical shifts when compared to ketones, typically in the
region of 220-280 ppm. 70-72
Thioamides also have higher chemical shifts relative to
45
C chemical shift of 210 ppm, by way of
example. However, even when 62 was subjected to wide 13
C NMR spectroscopy,
scanning between 0-300 ppm, the thioketone and the thioamide peaks were not
observed. Also worth noting, is that the methyl carbons in the tBu group responsible for
the signal at 27.5 ppm in the 13
C NMR spectrum is uncharacteristically broad. Single-
bond HMQC NMR experiments confirmed however that this broad peak was indeed
caused by the methyl carbon atoms in the tBu group coupling to the methyl hydrogens
at δ 1.12 ppm in the 1 H NMR (Figure 1.12).
Figure 1.12
Thus, self-addition product 63 could not be the unknown product 62. This was also
confirmed by the isolation of an authentic sample of 63 from other experiments which
will be discussed later. By way of comparison, the analytical data for the authentic self-
addition product 63 is presented in Figures 1.13, 1.14 and 1.15:
46
Data for 63:
1 H NMR (400 MHz, CDCl3) 6.09 (ddd, J = 9.5, 4.5, 1.5 Hz, 1H, CHN), 4.59–4.51 (m,
1H, CHAHBN), 4.47 (ddd, J = 9.5, 9.5, 5.0 Hz, 1H, CHAHBN), 2.59–2.49 (m, 1H,
CHAHB), 1.88–1.79 (m, 1H, CHAHB), 1.39 (s, 9H, CMe3), 1.35 (s, 9H, CMe3)
13 C NMR (100.6 MHz, CDCl3) δ 259.2 (C=S), 209.2 (NC=S), 74.8 (CHN), 55.0
(CH2N), 51.7 (CMe3), 43.1 (CMe3), 30.1 (CMe3), 29.8 (CMe3), 24.4 (CH2).
Hitherto however, we have been unable to conclusively determine the identity of
product 62. Scheme 1.39 represents our best suggestion for the structure of unknown
compound 62, as well as the mechanism for its formation. Initial attack of lithiated
azetidine 64 on the sulfur atom of the thioamide 35 forms a second organolithium,
partially stabilised by the electron withdrawing effect of the nitrogen atom and the α-
anion stabilising effect of the sulfur atom. Intramolecular attack of the organolithium on
the thioamide group then forms the bicyclic intermediate, followed by expulsion of
azetidine upon quenching with methanol. Finally, the sulfide attacks the thioxonium ion
to generate the final product 62.
48
Scheme 1.39
As we expected that solvent and temperature would affect the formation of the unknown
product 62, we decided to repeat the reaction at –100 °C, reduce lithiation times and
change the solvent from THF to Et2O. Thus, azetidine 35 was lithiated in Et2O at –100
°C for 2 minutes. In order to determine whether complete lithiation occurs under these
conditions, CD3OD was used as the electrophile rather than MeOH (Scheme 1.40).
Under these conditions, starting deuterated azetidine 65 was recovered in 90% yield,
with 100% deuteration occuring, as well as 14% of self-addition product rac-63. The
greater than quantitative recovery of starting material was possibly due to a minor
amount of contamination from long chain hydrocarbons that occurred during column
chromatography.
Scheme 1.40
13 C NMR spectroscopy of 63 showed the expected thioketone and thioamide peaks at
259 and 210 ppm respectively. 1 H NMR spectroscopy showed 100% deuteration α to
the nitrogen atom for 65 and no deuterium incorporation for self-addition product rac-
63. The percentage of deuterium incorporation can be calculated by taking the ratio of
the integration value between the CH2 group and the adjacent CH2N/CHDN groups
49
(Figure 1.16). At 0% deuteration, each CH2N group, as well as the CH2 group, will
integrate to 2H. At 100% deuteration, each CH2N/CHDN group will integrate to 1.5,
whereas the CH2 group will remain unchanged at 2H. Thus, the observed ratio that can
be seen from Figure 1.16 is 1.45:2, implying complete deuterium incorporation. 13
C – D
C NMR spectrum, with a coupling
constant of 23 Hz (Figure 1.17).
Figure 1.16
Figure 1.17
These results allowed several conclusions to be drawn. First, azetidine 35 is very easily
deprotonated, even at –100 °C. Second, that due to higher recovered yields of the
starting material, the extent to which side reactions can compete is substantially
lessened. Effectively all material was accounted for and the yield of recovered starting
material was almost doubled. Third, the tetrahedral intermediate formed by reaction of
35 with 64 is most likely to be stable under the reaction conditions until the addition of
CD3OD. If the tetrahedral intermediate were to collapse under the reaction conditions,
deprotonation of the acidic proton α to the thioketone would occur to generate a
thioenolate (Scheme 1.41). Given that deuteration of thioenolate 66 to form deuterated
self-addition product rac-67 is not observed, this reaction pathway does not occur. In
terms of pKa values, it is also highly unlikely that a sulfur anion (pKa ~10.5) would
expel a lithium amide (pKa ~35).
51
Scheme 1.41
Given as we had identified suitable conditions for the lithiation-trapping of azetidine 35,
the reactions with benzaldehyde and acrolein were carried out under the new lithiation
conditions, the results of which are shown in scheme 1.42, along with a tentative
explanation of the diastereoselectivity of the reaction: initial coordination of the oxygen
atom to the lithium ion serves to tether the electrophile to the organolithium reagent.
Trapping could then occur via a 2+2 cycloaddition-type mechanism. The lowest energy
configuration shown in scheme 1.42 maximises the distance between the bulky R and
tBu groups, thereby minimising steric hindrance and lowering the kinetic energy barrier
to reaction via this configuration.
Scheme 1.42
52
With benzaldehyde as electrophile, the total yield increased from 56% to 88% and, with
acrolein, the total yield almost doubled from 38% to 65%. The diastereoselectivity for
both reactions appeared to be relatively unchanged when compared to the results in
THF at –78 °C. Of the two electrophiles, acrolein is most chemically similar to
protected aldedyde 51, required for the total synthesis of (+)-retronecine. Yields for
acrolein were acceptable and the major diastereoisomer is the same as that required for
(+)-retronecine (+)-49, though the diastereoselectivity was lower than desired.
Using our optimised procedure, the reaction was tested on a variety of electrophiles to
check for generality and applicability. The electrophiles tested were CO2, Me2CO,
Ph2CO (Scheme 1.43), Me3SiCl (Scheme 1.44) and MeOCOCl (Scheme 1.45). Yields
for CO2, Ph2CO and Me2CO were much lower than the yields observed when
benzaldehyde or acrolein were used as electrophiles. Benzophenone adduct rac-68 was
insoluble in all but halogenated solvents and material was likely lost during flash
column chromatography in its purification. Propanone contains enolisable protons,
deprotonation of which by lithiated azetidine 64 could have contributed to the low
yields observed in this instance.
Scheme 1.43
Lithiation-trapping of azetidine 35 with methyl chloroformate as electrophile gave
methyl ester rac-70 in a disappointing 13% yield, with 62% yield of an additional
unknown product. HRMS gave an m/z value of 363.1534 (M+Na) + , which would be
consistent with the self-addition products meso/rac-71/72 shown in Scheme 1.44. The
1 H NMR,
13 C NMR and IR spectra gave results that were consistent with the proposed
53
structures for meso/rac-71/72, with the ketone being a key diagnostic peak in the IR
spectrum (1703 cm -1
C NMR spectrum (δ 204 and 202 ppm). Whilst the
relative stereochemistry of the major isomer could not be confirmed, the
diastereoselectivity could be ascertained by taking the ratio of integrals for the CHN
protons at δ 5.35 and 5.15 ppm in the 1 H NMR spectrum of the crude reaction mixture.
The 1 H NMR and
13 C NMR spectra of the purified 81:19 diastereomeric mixture of
meso/rac-71/72 are shown below in Figures 1.18 and 1.19.
Scheme 1.44
55
Lithiation trapping of azetidine 35 with Me3SiCl gave the expected α-silyl azetidine
rac-36 in 64% yield, as well as 14% of an unknown product with an m/z value of
307.1785 (M+H) + , indicating the presence of two Me3Si substitutions. NMR data, as
well as previous literature studies which will be discussed later, confirm that the
unknown product is doubly substituted 1,1-disilyl azetidine 73. This reaction and the
mechanism of its formation is shown below in Scheme 1.45.
Scheme 1.45
Theoretically, there are two additional isomers of disilane 73 that could form, cis-1,3-
disilane 74 and trans-1,3-disilane 75 (Scheme 1.46). Hodgson 58
demonstrated in 2015
that the half-life for interconversion of rotamers of azetidine 35 is 13.7 years at –78 °C!
As previously discussed, the lone pair on the sulfur atom will direct lithiation to the
same side of the azetidine ring. As the reaction temperature is –100 °C, rotameric
interconversion followed by lithiation at the 3-position is highly unlikely. Not only that,
but the initial silyl substitution serves to acidify the carbon atom it is bonded to via the
α-anion effect, making deprotonation at the 3-position even less likely.
Scheme 1.46
56
The 1 H NMR spectrum also confirmed that substitution occured to give the 1,1-disilyl
product 73 (Figure 1.20). Interconversion of rotamers in the product is likely to be slow
on the NMR timescale. Therefore the CHN signals for both 1,3-disilyl azetidines cis-74
and trans-75 would be non-equivalent and each would give its own signal. As can be
seen from the 1 H NMR spectrum of the obtained product (Figure 1.20), only 1 CH2N
signal is observed, confirming its identity as 1,1-disilyl azetidine 73. The C(SiMe3)2
atom in the 13
C NMR spectrum also appears as a quarternary centre, further confirming
the assignment of 73.
In summary, we have carefully studied the racemic lithiation-trapping of azetidine 35.
In particular, new conditions for the lithiation of 35 (sBuLi, TMEDA, Et2O, –100 °C, 2
min) were optimised. With some electrophiles and conditions, various side products
were isolated and where possible, characterised.
57
and Pyrrolidine
During this time, results obtained within our group 74
casted doubt on the viability of
completing an asymmetric synthesis of (+)-retronecine via the α-lithiation-trapping
methodology. The asymmetric α-lithiation-substitition reaction of azetidine 35 with
benzaldehyde was investigated by another member of our group using three separate
chiral diamines: (–)-sparteine (–)-1, Alexakis' diamine (S,S)-12 and the (+)-sparteine
surrogate (+)-13 (Table 1.8). Whilst the yields for the major diastereoisomer were
comparable, use of (S,S)-12 and (+)-13 gave effectively racemic product (Table 1.8,
entries 2 and 3). The use of (–)-sparteine (–)-1 as the chiral diamine gave the highest
enantioselectivities, though with the opposite configuration to that required for a
synthesis of (+)-retronecine.
Table 1.8: Role of Chiral Diamine in the Lithiation-Trapping of 35 with Benzaldehyde
58
Other recent results from within the group included the fact that lithiated azetidine 64
was configurationally unstable at –78 °C. Enantioenriched stannane (S)-37 which had
been prepared by an asymmetric lithiation-trapping with (–)-sparteine (–)-1 was
subjected to tin-lithium exchange using nBuLi and TMEDA in Et2O at –78 °C, followed
by trapping with benzaldehyde. The benzaldehyde adducts obtained were racemic
(Scheme 1.47), thus demonstrating facile interconversion between the enantiomers of
lithiated azetidine 64 at –78 °C as lithiated azetidine 64 was left for only 5 minutes
before electrophilic trapping.
Scheme 1.47
This result is in stark contrast to N-Boc pyrrolidine 7, N-Boc piperidine 17 and borane-
aziridine complex 27, which lithiate to give configurationally stable intermediates at –
78 °C. The lithiated species of the corresponding 5-membered analogue, N-thiopivaloyl
pyrrolidine 76, was also found to be configurationally unstable at –78 °C. An analogous
reaction was performed starting from enantioenriched stannane (S)-77, followed by tin-
lithium exchange and trapping with benzaldehyde (Scheme 1.48). Given that lithiated
N-Boc pyrrolidine is configurationally stable at –78 °C, it can be concluded that the
thiopivalamide directing group is responsible for the configurational instability in the 5-
and 4-membered rings.
Scheme 1.48
The conclusion from the previous work in the group is that the asymmetric lithiation-
trapping of azetidine 35 carried out by Hodgson and us cannot be an enantioselective
deprotonation, as this is not possible with configurationally unstable organolithium
species. This explains why (–)-sparteine (–)-1 and the (+)-sparteine surrogate (+)-13
59
give such different results (Table 1.8). The mechanism that accounts for
enantioinduction must therefore be either a dynamic thermodynamic resolution (Scheme
1.59) or a dynamic kinetic resolution (Scheme 1.60).
1.5.2 Electrophile Variation in the Asymmetric Lithiation-Substitution of N-
Thiopivaloyl Azetidine and Pyrrolidine
With a lack of precedent for high enantioselectivities in the correct sense for the
lithiation–trapping of azetidine 35, our attention turned away from the planned total
synthesis of (+)-retronecine. Instead, research focused on varying the electrophile and
diamine in the asymmetric lithiation-trapping reactions, with an ultimate aim to fully
explore the scope and limitations of this chemistry. In addition, we wished to clarify the
sense of induction with as many electrophiles as possible, as the group’s initial results
indicated that different electrophiles were giving different absolute configurations under
identical conditions for lithiation (Scheme 1.49). Hence, much time and effort was spent
developing approaches to proving the absolute configuration of the major products.
Scheme 1.49
As outlined above, lithiation-trapping of azetidine 35 with benzaldehyde and methyl
iodide gave products with different absolute stereochemistry, but gave products with the
same configuration were obtained for the corresponding pyrrolidine 76. Therefore, the
reaction was repeated using other electrophiles.
To begin with, azetidine 35 was subjected to an asymmetric lithiation-trapping with
sBuLi/(–)-1 in Et2O at –78 °C for 30 minutes, then trapped with carbon dioxide to give
the carboxylic acid (S)-69 in 96% yield and 75:25 er (Scheme 1.50). Interestingly,
under these conditions, self-addition products 62 and 63 were not observed in the 1 H
NMR spectrum of the crude reaction mixture. As previously stated, (–)-sparteine (–)-1
is known to be a much less activating diamine than TMEDA. Not only that, TMEDA is
sterically much smaller than (–)-1. It is feasible to imagine that the additional steric bulk
of (–)-1 prevents the reactive carbon centre of lithiated azetidine 64 from getting into
close proximity to the C=S group to form the self-addition product.
Scheme 1.50
The enantioselectivities of the final product (S)-69 was determined by taking a small
quantity (~ 2 mg) of the solid product (S)-69, forming the methyl ester by treatment
with TMS-diazomethane in toluene/MeOH 75, 76
and then subjecting the product ester to
chiral stationary phase HPLC. It was noted that when the reaction was repeated under
identical conditions, (S)-69 was found to have an er of 89:11. When the entirety of both
samples were re-dissolved, an aliquot of both solutions taken then separately derivatized
to the methyl ester, both samples showed an identical enantiomeric ratio of 75:25. This
result led to the realisation that (S)-69 can enantioenrich via crystallisation. We
wondered whether we could obtain a highly enantioenriched sample suitable for X-ray
diffraction by recrystallisation. Indeed, a single recrystallisation of (S)-69 with an er of
75:25 from hexane/EtOAc allowed isolation of (S)-69 in with an er of 97:3 from the
filtrate (ie,. racemic crystals of 69 crystallised). From this material, single crystals
61
suitable for X-ray diffraction were grown and an X-ray structure was obtained,
confirming the absolute stereochemistry of (S)-69 (Figure 1.21).
Figure 1.21
Given that the configuration of acid (S)-69 is opposite to the configuration α to nitrogen
of benzaldehyde adducts (R,R)-38 and (R,S)-57 α to the nitrogen, we desired further
proof of stereochemistry of acid (S)-69. Therefore, an independent synthesis from
commercially available, enantiomerically pure (S)-azetidine carboxylic acid (S)-82 was
developed. Enantiomerically pure (S)-82 was esterified to the methyl ester using TMS-
diazomethane in toluene/MeOH, the Boc group removed via treatment with TFA and
the nitrogen acylated with pivaloyl chloride to give enantiomerically pure (S)-83 in an
overall yield of 41% over 3 steps (Scheme 1.51). Carboxylic acid (S)-69 with an er of
75:25, prepared via sBuLi/(–)-1 lithiation-trapping of azetidine 35, was oxidised to the
amide by treatment with oxone in a 1:1 mixture of water/MeOH, then the crude residues
esterified to the methyl ester with TMS-diazomethane to give (S)-83 in a 47% yield
over 2 steps and an er of 76:24. A racemic sample of rac-83 was also prepared via the
same method from racemic carboxylic acid 69. Chiral stationary phase HPLC analysis
was carried out on rac-83 to find appropriate conditions to separate the two enantiomers.
Then, enantiomerically pure (S)-83 was subjected to chiral stationary phase HPLC to
determine which of the two peaks in the HPLC trace corresponded to the (S)-enantiomer.
Finally, (S)-83 of a 76:24 er was subjected to chiral stationary phase HPLC to confirm
that the major isomer was the (S) enantiomer.
62
Scheme 1.51
We wanted to investigate whether the corresponding pyrrolidine 76 would react with
CO2 to give the corresponding carboxylic acid with the opposite configuration to the
products obtained by reaction with benzaldehyde. Pyrrolidine 76 was lithiated using
sBuLi/(–)-1 in Et2O at –78 °C for 1 h, then trapped with CO2 to give acid (S)-87 in 76%
yield and an er of 80:20 (Scheme 1.52).
Scheme 1.52
Initial proof of stereochemistry followed the route of the previously established method
for (S)-69. Carboxylic acid (S)-87 was enantioenriched via recrystallisation from
hexane/EtOAc, with three successive recrystallisations improving the er from 80:20 to
90:10, to 96:4 and finally to 98:2. Single crystals were then grown from this
enantioenriched material and submitted for X-ray diffraction, confirming the
stereochemical assignment of (S)-87 (Figure 1.22).
63
As with the corresponding azetidine acid (S)-69, a two-step oxidation-esterification
procedure was used to form (S)-88 in quantitative yield and 98:2 er. Enantiomerically
pure (S)-proline was subjected to a Fischer esterification 77
to form the methyl ester (S)-
90, then acylated with pivaloyl chloride to form enantiomerically pure (S)-88 in a 69%
yield over two steps (Scheme 1.53). A racemic sample of rac-88 was also prepared
using the same two step oxidation-esterification procedure using rac-87 as starting
material. HPLC analysis of the racemate, the enantiomerically pure (S)-87 and (S)-87 of
98:2 er confirmed the original X-ray assignment.
Scheme 1.53
With carbon dioxide as an electrophile having been fully explored, our attention turned
to the use of methyl chloroformate as the electrophile. Lithiation of azetidine 35 with
sBuLi/(–)-1 in Et2O at –78 °C for 30 minutes, followed by trapping with methyl
chloroformate gave methyl ester (R)-70 in a 45% yield and 67:33 er, as well as 26% of
64
self-addition products 71/72 in an 80:20 diastereomeric ratio (Scheme 1.54). Proving
the absolute stereochemistry of (R)-70 was trivial, as acid (S)-69 with known
stereochemistry had to be derivatized to the corresponding methyl ester (S)-70 prior to
HPLC analysis.
Scheme 1.54
For the corresponding reaction on pyrrolidine 76, lithiation with sBuLi/(–)-1 in Et2O
at –78 °C for 1 hour, followed by trapping with methyl chloroformate gave (R)-91 in
58% yield and a 76:24 er (Scheme 1.55). Curiously, the analogous self-addition
products seen in the azetidine system was not observed in the pyrrolidine ring. Proving
the absolute stereochemistry was straightforward due to the known stereochemistry of
acid (S)-87.
Scheme 1.55
The use of methyl chloroformate as the electrophile gives products with the same
configuration α to the nitrogen atom as to those obtained with benzaldehyde as the
electrophile. Carbon dioxide, however, appears to react with lithiated azetidine 64 and
pyrrolidine 78 with inversion of stereochemistry.
Other electrophiles including benzophenone, acetone and Me3SiCl were also used in the
sBuLi/(–)-1 lithiation-trapping of azetidine 35. However, enantioselectivities in all
instances were poor and the absolute stereochemistry of the adducts were not
determined (Scheme 1.56)
Scheme 1.56
We also briefly investigated the use of (S,S)-Alexakis' diamine (S,S)-12 as a substitute
chiral diamine for (–)-1. Azetidine 35 was lithiated using sBuLi/(S,S)-12 in Et2O at –78
°C for 30 minutes, then trapped using Me3SiCl, carbon dioxide and methyl iodide (to
check reproducibility with Hodgson's results) (Scheme 1.57). The highest
enantioselectivities were achieved with the comparatively unreactive electrophile
methyl iodide, with carbon dioxide and Me3SiCl giving almost racemic products.
Scheme 1.57
With enough results in hand, we then attempted to account for the mechanism of
enantioinduction in the asymmetric lithiation-trapping of azetidine 35 and pyrrolidine
76. An overview of the most important results is shown in Scheme 1.58.
66
Scheme 1.58
The sense of induction in the final products is cleaerly dependent upon the electrophile
used. Use of carbon dioxide gives a product of opposite configuration to methyl
chloroformate and benzaldehyde. To the best of our knowledge, this is the first example
of a non-benzylic C–Li stereocentre reacting with inversion of stereochemistry. Use of
methyl iodide as electrophile also gives products of opposite configuration between the
4- and 5- membered rings. Enantioselectivities appear to decrease as the reactivity of the
electrophile decreases; benzaldehyde and carbon dioxide give near identical
enantioselectivities, methyl chloroformate slightly lower and methyl iodide the lowest.
One mechanistic scenario for fast trapping electrophiles such as benzaldehyde, carbon
dioxide and methyl chloroformate is shown in Scheme 1.59. At –78 °C, a
thermodynamic equilibrium consisting of a 75:25 mixture of diastereomeric lithiated
67
azetidines is established. With reactive electrophiles, the rate of reaction of each
diastereomeric organolithium is faster than the rate of interconversion between the
diastereomeric lithiated azetidines. Due to this, the enantiomeric ratio of the products is
the same as, or close to, the ratio of diastereoisomers of lithiated azetidines. With less
reactive electrophiles such as methyl iodide, the rate of reaction between each of the
lithiated azetidines with the electrophile is slower than the rate of interconversion
between the diastereoisomers of lithiated azetidine. Due to this, the enantioselectivities
in the final product are dependent upon the different rates of reaction between each
diastereomeric lithiated azetidine complex and the electrophile. The difference in the
sense of induction between methylated products (R)-40 and (S)-81 could be due to the
(R)-64 lithiated azetidine and (S)-78 lithiated pyrrolidine being the more reactive
diastereomeric complexes.
Scheme 1.59
At –78°C, (–)-1 appears to favour the (S) enantiomer of lithiated azetidine 64 in a 75:25
ratio and the (S) enantiomer of lithiated pyrrolidine 78 in a 80:20 ratio. This is too low
to realistically be synthetically useful.
In contrast to (–)-1, (S,S)-12 as the chiral diamine appears to give the opposite trend,
whereby best enantioselectivities are achieved with methyl iodide, the least reactive
electrophile. Carbon dioxide as the electrophile gives (S)-69 in a 57:43 er, implying that
there is a 57:43 thermodynamic ratio between diastereoisomers of lithiated azetidines.
68
As methyl iodide gives a higher enantioselectivity than carbon dioxide, it would suggest
that enantioinduction is due, in this case, to a dynamic kinetic resolution, whereby the
rate of reaction between diastereomeric lithiated azetidines is slower than their rate of
interconversion (Scheme 1.60).
1.6 Conclusions and Future Work
Azetidine 35 has been shown to lithiate rapidly using sBuLi/TMEDA in Et2O, even at –
100 °C, to form a highly reactive TMEDA/lithiated azetidine complex, as demonstrated
by deuteration incorporation experiments. Self-addition has been observed as a
deletrious side reaction under these racemic conditions, with the isolation of 62 and 63
being proof of this. An optimised procedure for the racemic lithiation-trapping of
azetidine 35 has been developed, allowing yields as high as 88% when benzaldehyde is
used as the electrophile. Diastereoselectivity for the two aldehyde products was shown
to be modest, although the relative stereochemistry of the major diastereoisomer was
shown to be correct for our planned synthesis of (+)-retronecine. The difference in the
sense of induction of the final products when carbon dioxide and methyl chloroformate
are used in the (–)-1 mediated lithiation-trapping of 35 and 76 is an unexpected result.
(Scheme 1.61).
Scheme 1.61
Lithiated azetidine 64 and pyrrolidine 78 are configurationally unstable at –78 °C. The
use of (–)-1 as a chiral diamine favours the formation of the (S) enantiomer of lithiated
azetidine 64 and pyrrolidine 78 in a 75:25 and 80:20 ratio respectively. Reactive
electrophiles are thus most likely to retain this ratio in the final products. (S,S)-12
appears to induce enantioselectivity via a dynamic kinetic resolution, given that CO2
gives a product with a lower enantiomeric ratio than with methyl iodide. Overall, the
mechanism of enantioinduction is dependent upon both the ligand and the electrophile
employed in the reaction. The poor enantioselectivities observed with the (+)-sparteine
surrogate (+)-13 and (S,S)-Alexakis' diamine (S,S)-12 have meant that the asymmetric
synthesis of (+)-retronecine was not feasible. Future work could include the use of the
Botc group, recently reported by Hodgson, 57
as well as a variety of other chiral
diamines to find appropriate conditions to allow an asymmetric total synthesis of (+)-
retronecine to be attempted (Scheme 1.62).
70
2.1 Introduction to Antibiotic Resistance
One of the most revolutionary developments in human history has been the discovery of
antibiotics; injuries that would have previously been considered a death sentence are
now treated as a matter of routine. Nonetheless, as antibiotic use has become more
frequent and widespread, so too have resistant strains of bacteria. The centre for disease
control (CDC) lists 18 different antibiotic resistant bacteria of varying threat levels,
including multi-drug resistant Neisseria gonorrhoeae, methicillin-resistant
Staphylococcus aureus and alarmingly, vancomycin – an antibiotic of last resort –
resistant enteroccoci and Staphylococcus aureus. Indeed, in 2014, the World Health
Organisation stated that: 78
"Antimicrobial resistance (AMR) within a wide range of infectious agents is a
growing public health threat of broad concern to countries and multiple sectors.
Increasingly, governments around the world are beginning to pay attention to a
problem so serious that it threatens the achievements of modern medicine. A
post-antibiotic era – in which common infections and minor injuries can kill –
far from being an apocalyptic fantasy, is instead a very real possibility for the
21 st century."
The problem of antibiotic resistant bacteria is compounded by the fact that there have
only been two classes of new synthetic antibiotics discovered within the past sixty
years, 79
the fluoroquinolones in 1961 and oxazolodinones in 1955, of which, only the
former possess broad spectrum antibiotic activity. Similarly, for natural products with
antibiotic activity, there have been only three that have been approved for medical use
within the last forty years – daptomycin, quinupristin–dalfopristin and fidaxomicin –
and each was deemed to be inappropriate for medical use when first discovered. Suffice
to say, the need for new classes of antibiotics is urgent and difficult to overemphasise.
72
2.2 Introduction to Anthracimycin And Chlorotonil
In 2013, Fenical isolated anthracimycin 92 from a strain of Streptomyces found in a sea
sponge off the coast of California. 80, 81
Anthracimycin was found to possess significant
activity towards Bacillus anthracis, methicillin-resistant Staphylococcus aureus and
vancomycin-resistant Enterococci, 82
as well as a range of other gram positive bacteria.
Interestingly, a similar product, chlorotonil 93, was found in an unrelated bacterium,
Sorangium cellulosum and was recently discovered to possess potent antimalarial
activity. 83, 84
Chlorotonil 93 has an additional methyl group on the decalin ring and a
gem-dichloro group. Perhaps most interesting, however, is that is has the opposite
configuration to anthracimycin 92 at all remaining stereocentres.
Figure 2.1
Given the structural similarities between anthracimycin 92 and chlorotonil 93,
anthracimycin was doubly chlorinated with N-chlorosuccinimide to give dichloro–
anthracimycin 94 and its biological activity tested and compared to anthracimycin.
(Table 2.1). 80
Table 2.1: Biological Activity Comparison Between 92 and 94
Anthracimycin 92 showed potent antibiotic activity towards gram positive bacteria B.
anthracis, S. aureus and E. faecalis, yet poor activity towards gram negative bacteria E.
coli, H. influenzae, B. thailandensis and P. aeruginosa, whereas dichloro-anthracimycin
94 is significantly more potent towards the gram negative bacteria with slightly
decreased potency towards gram positive bacteria. It is speculated that dichlorination
allows dichloro-anthracimycin 94 to cross the cell barrier more easily, although
currently this remains unproven.
Anthracimycin 92, unlike most other macrocyclic antibiobiotics, does not exert its
biological effects by inhibiting protein synthesis by binding to the 50S ribosomal
subunit. In contrast, it interferes with DNA replication, although the specific mechanism
by which it does this has yet to be confirmed. 82
Recently, the gene cluster responsible
for the biosynthesis of anthracimycin 92 was identified and the enzyme-mediated
biosynthetic pathway is shown in Scheme 2.1. Starting from malonyl coenzyme A 95,
the biosynthetic pathway consists of multiple chain elongation reactions, with the key
step being the intramolecular Diels-Alder reaction to form the decalin 96 and setting up
the correct stereochemistry of four stereocentres.
74
Scheme 2.1
The structurally related chlorotonil 93 on the other hand, is toxic towards all stages of
Plasmodium falciparum, the organism responsible for causing malaria, with an IC50 of
between 4-32 nm, depending on strain. The fact that chlorotonil 93 exhibits low toxicity
towards mammalian cells makes it an attractive target for medicinal use.
75
Remarkably, a total synthesis of chlorotonil 93 was reported in 2008, the same year that
the natural product was isolated by Kalesse (Schemes 2.2-2.6). 85
Fragment 97 was
reaction and
followed by a Corey-Fuchs homologation 89
to give 97 in six steps in an overall yield of
41% (Scheme 2.2) .
between PMB-protected alcohol 101 and vinyl boronic ester
102 gave fragment 103 in 81% yield and as a single geometric isomer. Suzuki
coupling 91, 92
of 97 with boronic ester 103 gave brominated 104 in 76% yield which was
then deprotected, oxidised and subjected to a stabilised Wittig reaction to give 105 in
70% yield over three steps. Finally, a one-pot intramolecular Lewis acid-mediated
Diels-Alder reaction, PMB deprotection and lactonisation gave the tricyclic core 106 in
58% yield and a 13:1 diastereomeric ratio (Scheme 2.3).
76
Scheme 2.3
Importantly, the presence of the vinylic bromine atom was necessary for the Diels-Alder
step to proceed in high diastereoselectivity. When the same reaction conditions were
applied to the debrominated derivative 107, diastereoisomers 108, 109 and 110 were
isolated in 49% overall yield and a greatly reduced 1:1:3 diastereoselectivity (Scheme
2.4).
77
Next, functionalised allyl phosphonate 111 was synthesised in another cross metathesis
reaction in 39% overall yield from allyl bromide (Scheme 2.5).
Scheme 2.5
Brominated fragment 106 was then dehalogenated by treatment with sodium amalgam
and the lactone opened with potassium hydroxide to give the intermittant carboxylic
acid. Then, methylation with TMS-diazomethane, 93
oxidation with DMP gave the
aldehyde 112 in a 70% yield over three steps. Next, a second Still-Gennari reaction
between 112 and allyl phosphonate 111 gave tetraene 113 in 55% yield and a 3:1 E/Z to
E/E ratio, followed by a Claisen condensation to give 114. This was then immediately
treated with BF3.OEt2 to deprotect the PMB ether and affect diastereoselective ring
closure to give 115 in 47% yield over two steps. Finally, di-chlorination with NCS gave
chlrotonil 93 in 65% yield. The complete synthesis proceeds in 18 linear steps and 1.5%
overall yield.
2.3 Project Outline
Anthracimycin 92 has a very interesting antibiotic profile and has yet to be made
synthetically. The development of a total synthesis would allow analogues to be
prepared and evaluated for biological activity as antimicrobial agents. Also, given the
structure-activity relationship between anthracimycin 92, di-chloroanthracimycin 94 and
chlorotonil 93, our aim was to develop a racemic synthesis of anthracimycin to allow
both enantiomers to be tested for biological activity. As such, all structures presented
hereafter will be racemic unless stated otherwise. Our proposed synthesis of the
macrocycle follows a similar route to that published by Kalesse for the synthesis of
chlorotonil 93, but includes a new approach to the synthesis of the decalin core 116 and
is shown below in Scheme 2.7. Most of our efforts focused on the early steps of this
synthesis.
Scheme 2.7
From the starting enone 117, a stereoselective one-pot 1,4-addition of allyl cuprate,
followed by trapping of the intermediate enolate with methallyl bromide is expected to
give 118. Ring closing metathesis, for example, with Grubbs 2 nd
generation catalyst
would then form the desired trans-decalin 119. Next, regioselective formation of the
80
less substituted enolate with LDA, trapping with phenylselenium bromide and oxidative
work-up was planned to then give enone 120. Stereoselective 1,2-reduction should then
give alcohol 121 which would then be acylated with proprionyl chloride to give the
propionyl ester 122. The resulting ester would be treated with LDA/DMPU to form the
(E)-enolate, 94
trapped with Me3SiCl and then warmed to allow a stereospecific Ireland-
Claisen 95
The relative stereochemistry of the five contiguous stereocentres in carboxylic acid 116
will be controlled in the following ways. First, cyclohexenones such as 117 adopt a
relatively flat orientation. 96
The initial addition of allyl cuprate to 117 should therefore
occur from the opposite face to the bulky CH2OTIPS substituent to minimize steric
repulsion (Scheme 2.8). 97
This will generate the intermediate enolate. As the allyl group
is now closer to the nucleophilic centre, the approach of methallyl bromide to the
enolate will occur opposite to it, thus creating diene 118 with all ring substituents
adopting equatorial positions.
Secondly, reduction of 120 using the sterically bulky LiAlH(OtBu)3 should proceed via
equatorial attack of hydride to give alcohol 121 with the hydroxyl and CH2OTIPS
groups on the same face (Scheme 2.9).
Scheme 2.9
Thirdly, proprionyl ester 122 will be treated with LDA/DMPU to selectively form the
(E)-enolate, 94
trapped with Me3SiCl to form the corresponding (Z)-silyl enol ether 123
81
rearrangement to give
carboxylic acid 116. The transition state of this reaction is shown in Scheme 2.10. The
configuration of the exocyclic methyl group in carboxylic acid 116 will be dependent
upon the enolate geometry. Should the Ireland-Claisen rearrangement proceed via a
different transition state and give the methyl group with the opposite configuration, the
corresponding (E)-silyl enol ether could be formed by deprotonation with LDA in the
absence of DMPU.
Scheme 2.10
The proposed end-game synthesis to take carboxylic acid 116 to anthracimycin 92 is
shown below in Scheme 2.11. LiAlH4 reduction of the carboxylic acid, DMP oxidation
to the aldehyde and a (Z)-selective Still-Gennari olefination using the PMB-protected
allyl phosphonate 124 should give the conjugated diene 125. Next, removal of the TIPS
group with hydrogen fluoride in pyridine, Jones oxidation 98
to the carboxylic acid
followed by methylation with TMS-diazomethane would give the corresponding methyl
ester 126. At this point, our synthesis follows that of Kalesse. 85
A Claisen condensation
of 126 with the dianion formed from ethylacetoacetate, PMB deprotection and
diastereselective ring closure by treatment with BF3.OEt2 should allow access to
anthracimycin 92. Anthracimycin 92 is a pseudo-enantiomer of chlorotonil 93, apart
from the additional methyl group on chlorotonil 93, anthracimycin 92 possesses the
opposite configuration at all remaining stereocentres. It would therefore be reasonable
82
to assume the ring closing step to form anthracimycin 92 from 127 will place the
malonic methyl group in the opposite configuration to chlorotonil 93.
Scheme 2.11
In this chapter, our efforts exploring the early parts of this proposed route to
anthracimycin 92 are described.
2.4.1 Preparation of the Starting Enone
As it is the beginning of a total synthesis, a reliable, scalable synthesis of enone 117 was
required. Cyclohexanones with substitution at the 4-position such as 128, 129, 130 and
131 are prohibitively expensive to be used as the starting material of a multi-step total
synthesis, their per-gram prices from Sigma-Aldrich being shown below in Figure 2.2.
Conveniently, the corresponding 1,4-disubstituted aromatic derivative 132, presents a
much more economically viable starting material for the synthesis of TIPS protected
enone 117. Thus, our proposed synthesis began with the hydrogenation of 4-
hydroxybenzyl alcohol 132 (Scheme 2.12).
Figure 2.2
With the chosen starting material in hand, our overall aim was to carry out ring
hydrogenation, followed by TIPS protection of the primary alcohol. This would be
followed by two successive oxidations to get to the desired TIPS protected enone 117
(Scheme 2.12).
Scheme 2.12
Initially, small scale reactions were carried out to find appropriate conditions for the
large scale synthesis of enone 117. Use of Pd/C as a heterogeneous catalyst led to
84
cleavage at the benzylic C–O position to give p-cresol as the major product. Changing
the catalyst to PtO2, known for hydrogenation of aromatic rings, 99
led to isolation of the
desired saturated diols 133/134 as a 1:2 inseparable mixture of stereoisomers (by 1 H
NMR spectroscopy) in a combined yield of 58%. The reaction was very slow, requiring
five days to reach completion. However, the catalyst loading was very low at 1 mol%
and from a practical perspective, very straightforwards, requiring only occasional
refreshing of the hydrogen source. The reaction also proved very amenable to scale up
due to the robustness of the chemistry and the low cost of starting materials. The results
of the multigram synthesis will be presented later.
Heterogenous hydrogenations of alkenes typically proceed from the least hindered
face. 100-105
For the purposes of this discussion, the cis isomer is assumed to be the major
isomer, although this was not definitively assigned due to the difficulty in assigning J
values for 133/134, with the 1 H NMR spectrum shown in Figure 2.3. The
stereoselectivity for the reaction was taken by the ratio of the integrals from the signals
at approximately δ 3.73 and 3.57 ppm, corresponding to the CH(OH) protons.
Figure 2.3
85
Diols 133/134 were then selectively protected at the primary alcohol by treatment with
TIPS-Cl in a total yield of 74% (Scheme 2.13).
Scheme 2.13
Initially, it was unclear whether or not one of stereoisomers trans-135 and cis-136 were
in fact potential regioisomers 137 or 138 (Scheme 2.13). The identity of 135 and 136
was confirmed by independently oxidising a pure sample of each isomer, each giving
139 with identical analytical data (Scheme 2.14).
Scheme 2.14
TIPS protected ketone 139 also proved amenable to a multigram scale synthesis via the
same route (Scheme 2.15). Yields for the hydrogenation of phenol 132 to give 133/134
could be improved to 77% with longer reaction times (7 days) and reduced catalyst
loadings (0.44 mol%) with no change in stereoisomeric ratio. Silylation of the primary
alcohol also proceeded in a similar fashion upon scale up, with yields of 82% being
observed with 2 g of diols 133/134 as starting material. Finally, ketone 139 could be
formed in 72% via a Swern oxidation 106-108
or an improved 82% via a Dess-Martin
periodinane oxidation. 109, 110
2.4.2 Oxidation of Ketone 139
There are several known methods for the introduction of alkene functionality into
ketones to generate the corresponding enones. Most commonly, methods for the
oxidation of ketones into enones include α-bromination and elimination, 111
sulfoxide/selenoxide elimination, 112, 113
and the
Not all these methods are equally suited for large
scale synthesis. Selenium reagents are both toxic and expensive and the formation of the
α-keto selenide intermediate requires large amounts of pyrophoric organolithium
reagents. Sulfides are typically formed via the same methodology and although the
sulfur reagents are usually less toxic, elimination of the sulfoxide requires elevated
temperatures which often leads to deletrious side reactions occurring. The Saegusa
oxidation, whereby silyl enol ethers are treated with palladium(II) acetate in the
presence of a stoichometric oxidant, often requires catalyst loadings in excess of 50
mol%, usually relegating it to late-stage synthesis. For these reasons, robust, scalable
methodology for the introduction of the alkene bond into ketone 139 needed to be
developed. In order not to waste ketone 139, commercially available 4-methyl
cyclohexenone 140 was used initially in its place.
87
To begin with, cyclohexenone 140 was subjected to a two-step bromination-elimination
reaction (Scheme 2.16). Bromination in CH2Cl2 proceeded rapidly at room temperature,
but bromo-ketones 141/142 were found to be unstable and decomposed at room
temperature, forming a dark, tarry material over several hours. Attempts to immediately
eliminate the crude bromo-ketones 141/142 without prior purification, however, failed
to give the desired enone 143.
Scheme 2.16
Also investigated was the use of a Rubottom oxidation of silyl enol ether 144, followed
by mesylation and elimination of the α-hydroxy ketones 145/146 (Scheme 2.17).
Cyclohexenone 140 was converted into the corresponding silyl enol ether 144 by
treatment with Me3SiCl, sodium iodide and triethylamine in acetonitrile in 72% yield
after distillation under reduced pressure. Then, reaction of 144 with buffered mCPBA,
followed by acid-catalysed epoxide opening successfully gave a 57:43 diastereomeric
mixture of α-hydroxy ketones 145/146 in 41% yield. Unfortunately, attempted
mesylation-elimination of 145/146 gave rise to a complex mixture of products.
Scheme 2.17
The most promising method attempted was a Saegusa oxidation of silyl enol ether 144
(Scheme 2.18). Although only 25 mg of crude product was obtained from 0.59 g of the
starting silyl enol ether 144, 1 H NMR spectroscopic data of the crude material showed
complete conversion into enone 143.
88
Scheme 2.18
The practical procedure for the isolat